Comment - FBXL Lotide

Can you really blame anyone who turns to AI, because that garbage at least sounds like it tries to help you?

A comfortable lie is still a lie. Everything that comes out of an LLM is a lie until proven otherwise. (“Lie” is a bit misleading, though, as they don’t have agency or intent: they’re a variation of your phone keyboard’s next-word text prediction algorithm. With added flattery and confidence.)

There’s a reason experienced people stress hard to others about not using them as shortcuts to your own knowledge. This is the outcome.

Another way to look at it is “trust, but verify”. If you’re intent on relying on probabilistic text as an answer, instead of bothering to learn, then take what it’s given you and verify what that does before doing it. You could learn to be an effective sloperator with just that common sense.

But if you’re going to give an LLM root/admin access to a production environment, then expect to be laughed at, because you had plenty of opportunities to not destroy something and actively chose not to use them.

source

Sort:hotnew top

irmadlad@lemmy.world ⁨5⁩ ⁨months⁩ ago

Everything that comes out of an LLM is a lie until proven otherwise.

Everything that comes off of a tutorial, or web page is paddling the same boat, without exception.

source
- lambalicious@lemmy.sdf.org ⁨5⁩ ⁨months⁩ ago
  Are you really comparing LLM output to be on the same level of… hallucination-ness, than a Gamefaqs tutorial for a SNES game from the late 90s?
  
  I know tiktok has deep-fried and rotten the brains of entire generations but this is just ridiculous.
  
  source
  - irmadlad@lemmy.world ⁨5⁩ ⁨months⁩ ago
    
    Gamefaqs tutorial for a SNES game
    
    Well, I can’t speak to Gamefaqs or SNES as I am incapable of gaming. However, even what I would consider reputable tutorials such as ones you find on HowToForge, sometimes don’t quite turn out as expected. There is always some config or app that the tutorial needed, but was left out by the person writing the tutorial. Or the writer of the tut, had something pre-configed or preinstalled, so it wasn’t mentioned, even when following the tutorial line by line. It’s inevitable. For this reason, I maintain a small test VPS where I can run amuck and if I fuck something up, no problem, wipe/reinstall. So, if you’re going to say don’t trust AI, then you have to also be skeptical of all tuts. I mean, that’s where the AI scrapers got the info in the first place. I’m not saying AI is 100% tho I anticipate one day it will be, or at least very damn close. There are things AI gets right. It seems very capable of writing compose files well. Just my 2p
    
    source
    lambalicious@lemmy.sdf.org ⁨5⁩ ⁨months⁩ ago
    
    However, even what I would consider reputable tutorials such as ones you find on HowToForge, sometimes don’t quite turn out as expected
    
    Yes, because that’s a natural process. Most tutorial s written by users cover the experience the user had in their own use case. They don’t / can’t cover the same ground or have access to the same levels of examination that the devs can have.
    
    So, if you’re going to say don’t trust AI, then you have to also be skeptical of all tuts. I mean, that’s where the AI scrapers got the info in the first place.
    
    Oh please. Stop licking corporate AI boot and drinking the kool-aid. There’s at least two orders of magnitude of truthfulness and trustability between “a discrete set of tutorials written to cover described use cases” and “a random mix and blend hodgepodge coke snort prisoner soup ectoplasm of all the above, fine-tuned to invent answers that produce gratification and brand dependence”. You saying that these two things are as trustable as each other suggests you have quite a misanthropic edge to your personality and/or are going through a stage of cult-of-personality (or cult-of-brand).
    
    I trust the humans who write the tutorials that have em-dashes. I don’t trust an AI that just slurped and pirated the work of those humans to try and snake-oil me with a bunch of grammar mistakes adorned with em-dashes.
    
    source
    -> View More Comments
squaresinger@lemmy.world ⁨5⁩ ⁨months⁩ ago
I had a problem with Fedora 42, where the performance of my games would be fine one day and abysmal another day. Couldn’t find a pattern. I googled a ton, tried to debug myself, asked on reddit, stackexchange, the fedora forum and lemmy. I only got answers like “Works fine on my machine, noob” and “I have that problem too”. It only affected games running in proton on heroic, everything else was fine.

After about a year of on-and-off debugging and asking around, I swallowed my pride and asked ChatGPT.

First answer from that thing was correct: I had run dnf update without doing a flatpak update right afterwards. Turns out, flatpak has its own copy of Nvidia drivers and if the system driver is updated without the flatpak copy being updated, it falls back to software rendering. So the performance was crap until I did flatpak update the next time, and broke again when I ran dnf update.

I still haven’t found that in any documentation so far.

AI is crap more often than not, but it does at least try to help and sometimes it actually does.

Look in this thread here. Is there even a single answer that tries to help OP, or is every single answer here just dumb snark?

source
- fodor@lemmy.zip ⁨5⁩ ⁨months⁩ ago
  I love how you ignore all the comments that actually tried to help OP. Or maybe you posted early on but you can’t be bothered to go back and edit your response. And then you pretend that AI tries to help us do anything, as if it had motivation or volition. Come on now.
  
  source
- Brewchin@lemmy.world ⁨5⁩ ⁨months⁩ ago
  It’s true that people on the internet can be dicks. Even more so technical people (and that’s not limited to online: those online dicks are usually IRL dicks when taking technical stuff). But that’s a hurdle, not a barrier.
  
  There’s little anyone here can do to help OP, as they (if I understand it correctly) have already irreparably nuked their hardware. The current problem is significantly different and harder than the original problem. Asking randos on this community is unlikely to yield results. Hence the focus on variations of “Now… what did we learn? 🤨”
  
  I’m not trying to help, as I’m not familiar enough with SAS nor the current problem. The same is likely true of others here.
  
  source
  - squaresinger@lemmy.world ⁨5⁩ ⁨months⁩ ago
    The only evidence that OP irreparably nuked their hardware is ChatGPTs word.
    
    The bigger issue ad hand is that everyone here is a noob when it comes to exotic hardware like SAS, and still everyone here thinks they are 1337 haxors enough to tell OP that they are a noob idiot.
    
    Tbh, if OP asks for help here and there’s 49 comments of people being dicks and one that actually helps it might be worthwhile. But as it stands it’s 50 to 0, with nobody here beeing educated enough to know anything about the subject.
    
    source